Re: [Help-glpk] A few questions

[Andrew Makhorin]

On Mon, 2018-05-28 at 12:13 -0500, STEVE VESTAL wrote:
> I have a couple of initial questions based on skimming the GLPK
> Reference Manual and examples.
> 
> 
> 
> The manual says set_glb_obj_coef applies to structural variables.  How
> are auxiliary (aka slack, row) variable costs set? 

Glpk does not support objective coefficients for auxiliary variables. If
you have a constraint, say, x1 + x2 + x3 >= 1, you may replace it with
equality x1 + x2 + x3 - t = 0 and bound inequality t >= 1, and assign
the obj coefficient to t.

> 
> 
> Are free variables implemented as the difference of two non-negative
> variables?  

No. For details please see Section 2.8 "Simplex method routines" and
Section 4.1 "Background" in the reference manual (glpk/doc/glpk.pdf).

> Is it possible to set the costs (objective coefficients) of the two to
> have the same sign so that the absolute value of a free variable can
> be maximized/minimized?

If needed, you may use equality x1 - x2 - f = 0, where f is a free
variable, x1 and x2 are non-negative variables.

> 
> 
> 
> Has anyone benchmarked GLPK against the netlib LP suite,
> http://www.netlib.org/lp/ 

Please see glpk/doc/netlib.txt .

> 
> 
> Are there any reports of how it has fared with those?  (The objective
> values obtained by rational solution have been published.)
> 
> 
> 
> Thanks.
> 
> _______________________________________________
> Help-glpk mailing list
> address@hidden
> https://lists.gnu.org/mailman/listinfo/help-glpk